[wwb]Take prompts from different categories for def dataset for VLM

sbalandi · sbalandi · commit 95a122d267a8 · 2025-06-25T17:19:12.000+01:00
diff --git a/tools/who_what_benchmark/whowhatbench/visualtext_evaluator.py b/tools/who_what_benchmark/whowhatbench/visualtext_evaluator.py
@@ -25,7 +25,7 @@ def prepare_default_data(num_samples=None):
     set_seed(42)
     default_dataset = datasets.load_dataset(
         DATASET_NAME, split="test", streaming=True
-    ).take(NUM_SAMPLES)
+    ).shuffle(42).take(NUM_SAMPLES)
     return default_dataset.map(
         lambda x: preprocess_fn(x), remove_columns=default_dataset.column_names
     )

Original file line number	Diff line number	Diff line change
`@@ -25,7 +25,7 @@ def prepare_default_data(num_samples=None):`
`25`	`25`	`set_seed(42)`
`26`	`26`	`default_dataset = datasets.load_dataset(`
`27`	`27`	`DATASET_NAME, split="test", streaming=True`
`28`		`- ).take(NUM_SAMPLES)`
	`28`	`+ ).shuffle(42).take(NUM_SAMPLES)`
`29`	`29`	`return default_dataset.map(`
`30`	`30`	`lambda x: preprocess_fn(x), remove_columns=default_dataset.column_names`
`31`	`31`	`)`